[SCIP] SCIP produces different results for the same problem instance

[Tufan BAYDEMİR]

Hi Gregor,
Thank you for your useful instruction.
When i tried with PYTHONHASHSEED=0, it generated exactly the same results.

Thank you all,
Regards

From: Scip <scip-bounces at zib.de> On Behalf Of Gregor Hendel
Sent: Wednesday, October 3, 2018 6:33 PM
To: scip at zib.de
Subject: Re: [SCIP] SCIP produces different results for the same problem instance

Dear Tufan,

we are actually making an effort to keep SCIP deterministic, so you should see the same solution after every call on the same problem. A time limit might lead to different results, but this does not seem to be the case here.

I tend to say you got "Pythoned" here.

My best guess, as I have no other information how you invoke the problem solve, is that you use dictionaries
somewhere in the problem creation. Now, depending on the version, Python's internal hashing is nondeterministic,
see also here<https://stackoverflow.com/questions/14956313/why-is-dictionary-ordering-non-deterministic>. The same would be true for Java, I suppose. If you loop over the keys() of a dictionary, the order
is different every time you run, effectively generating a different permutation of your problem's columns and rows, which may affect the performance of the MIP solver and also specific solution values may be different between runs, but of course not the objective value of an optimal solution.

The link I posted contains a link to a solution to the problem using a PYHASHSEED environment variable, which I haven't verified.

Let us know if this is the source of, and solution to, your problem,
Gregor

Am 03.10.2018 um 13:09 schrieb Tufan BAYDEMİR:
I’m using pyscipopt interface within Python to solve a MIP problem.
I run the same code without making any changes and obtain different optimal solutions.
The objective function value is the same for all solutions.
But value of the decision variables are always different ie. The program ganarates alternative optimal solutions in each subsequent run.

How can I insure that i get the same solution?

Regards
Tufan

obase Bilgisayar ve Danışmanlık Hizm. Tic. A.Ş.
TEL: +90216 527 30 00
FAX: +90216 527 31 11
[http://www.obase.com/images/signature/home.png]<http://www.obase.com>[http://www.obase.com/images/signature/facebook.png]<https://www.facebook.com/obasesocial>[http://www.obase.com/images/signature/twitter.png]<https://twitter.com/obasesocial>[http://www.obase.com/images/signature/linkedin.png]<https://tr.linkedin.com/in/obase>
[http://www.obase.com/images/signature/obaselogo.png]<http://www.obase.com>

Bu elektronik posta ve onunla iletilen bütün dosyalar sadece göndericisi tarafindan almasi amaclanan yetkili gercek ya da tüzel kisinin kullanimi icindir. Eger söz konusu yetkili alici degilseniz bu elektronik postanin icerigini aciklamaniz, kopyalamaniz, yönlendirmeniz ve kullanmaniz kesinlikle yasaktir ve bu elektronik postayi derhal silmeniz gerekmektedir. OBASE bu mesajin icerdigi bilgilerin doğruluğu veya eksiksiz oldugu konusunda herhangi bir garanti vermemektedir. Bu nedenle bu bilgilerin ne sekilde olursa olsun iceriginden, iletilmesinden, alinmasindan ve saklanmasindan sorumlu degildir. Bu mesajdaki görüsler yalnizca gönderen kisiye aittir ve OBASE görüslerini yansitmayabilir.

Bu e-posta bilinen bütün bilgisayar virüslerine karsi taranmistir.

This e-mail and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you are not the intended recipient you are hereby notified that any dissemination, forwarding, copying or use of any of the information is strictly prohibited, and the e-mail should immediately be deleted. OBASE makes no warranty as to the accuracy or completeness of any information contained in this message and hereby excludes any liability of any kind for the information contained therein or for the information transmission, recepxion, storage or use of such in any way whatsoever. The opinions expressed in this message belong to sender alone and may not necessarily reflect the opinions of OBASE.

This e-mail has been scanned for all known computer viruses.




_______________________________________________

Scip mailing list

Scip at zib.de<mailto:Scip at zib.de>

https://listserv.zib.de/mailman/listinfo/scip

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20181009/f408d592/attachment.html>